Vol 457 | 12 February 2009 | doi:10.

1038/nature07762

LETTERS
Metabolomic profiles delineate potential role for
sarcosine in prostate cancer progression
Arun Sreekumar1,2,3,4, Laila M. Poisson5*, Thekkelnaycke M. Rajendiran1,3*, Amjad P. Khan1,3*, Qi Cao1,3,
Jindan Yu1,3, Bharathi Laxman1,3, Rohit Mehra1,3, Robert J. Lonigro1,4, Yong Li1,3, Mukesh K. Nyati4,6, Aarif Ahsan6,
Shanker Kalyana-Sundaram1,3, Bo Han1,3, Xuhong Cao1,3, Jaeman Byun7, Gilbert S. Omenn2,7,8, Debashis Ghosh4,5,11,
Subramaniam Pennathur2,4,7, Danny C. Alexander12, Alvin Berger12, Jeffrey R. Shuster12, John T. Wei4,9,
Sooryanarayana Varambally1,3,4, Christopher Beecher1,2,3 & Arul M. Chinnaiyan1,2,3,4,9,10

Multiple, complex molecular events characterize cancer develop-
ment and progression1,2. Deciphering the molecular networks that
distinguish organ-confined disease from metastatic disease may
lead to the identification of critical biomarkers for cancer invasion
and disease aggressiveness. Although gene and protein expression
have been extensively profiled in human tumours, little is known
about the global metabolomic alterations that characterize neo-
plastic progression. Using a combination of high-throughput
liquid-and-gas-chromatography-based mass spectrometry, we
profiled more than 1,126 metabolites across 262 clinical samples
related to prostate cancer (42 tissues and 110 each of urine and
plasma). These unbiased metabolomic profiles were able to distin-
guish benign prostate, clinically localized prostate cancer and
metastatic disease. Sarcosine, an N-methyl derivative of the amino
acid glycine, was identified as a differential metabolite that was
highly increased during prostate cancer progression to metastasis
and can be detected non-invasively in urine. Sarcosine levels were
also increased in invasive prostate cancer cell lines relative to
benign prostate epithelial cells. Knockdown of glycine-N-methyl
transferase, the enzyme that generates sarcosine from glycine, atte-
nuated prostate cancer invasion. Addition of exogenous sarcosine
or knockdown of the enzyme that leads to sarcosine degradation,
sarcosine dehydrogenase, induced an invasive phenotype in benign
prostate epithelial cells. Androgen receptor and the ERG gene
fusion product coordinately regulate components of the sarcosine
pathway. Here, by profiling the metabolomic alterations of prostate
cancer progression, we reveal sarcosine as a potentially important
metabolic intermediary of cancer cell invasion and aggressivity.
To profile the ‘metabolome’ during prostate cancer progression, we
used both liquid and gas chromatography coupled with mass spec-
trometry3 to interrogate the relative levels of metabolites across 262
prostate-related biospecimens (outlined in Supplementary Fig. 1).
Specifically, 42 tissue samples and 110 matched specimens of plasma
and post-digital-rectal-exam urine from biopsy-positive cancer
patients (n 5 59) and biopsy-negative control individuals (n 5 51)
were assayed (Fig. 1a). A total of 1,126 metabolites were quantified
and, as expected, only a small percentage of these metabolites (15.6%)
were shared across the disparate biospecimen types (Fig. 1a).
Evaluation of the unbiased metabolomic profiles of plasma or
urine did not identify robust differences between biopsy-positive
and -negative individuals. For plasma, 20 out of 478 (4%) metabolites
1
The Michigan Center for Translational Pathology, 2Center for Computational Medicine and Biology, 3Department of Pathology, 4The Comprehensive Cancer Center, 5Department of
Biostatistics, 6Department of Radiation Oncology, 7Department of Internal Medicine, 8Department of Human Genetics, 9Department of Urology, 10Howard Hughes Medical Institute,
University of Michigan Medical School, Ann Arbor, Michigan 48109, USA. 11Department of Statistics and Huck Institute of Life Sciences, Penn State University, Pennsylvania 16802,
USA. 12Metabolon, Inc. 800 Capitola Drive, Suite 1 Durham, North Carolina 27713, USA.
*These authors contributed equally to this work.
910
©2009 Macmillan Publishers Limited. All rights reserved
were differential (Wilcoxon P , 0.05), with a false discovery rate
(FDR) of 99%. Likewise, for urine, 36 out of 583 (6%) metabolites
were differential (Wilcoxon P , 0.05), with an FDR of 67%. Thus, our
initial focus was directed towards understanding the tissue metabo-
lomic profiles as they showed more robust alterations.
Tissue samples were derived from benign adjacent prostate
(n 5 16), clinically localized prostate cancer (n 5 12, PCA) and meta-
static prostate cancer (n 5 14) patients. Selection of metastatic tissue
samples from different sites (see Supplementary Table 2) minimized
characterization of analytes specific to cells of non-prostatic origin. In
total, high-throughput profiling of the tissues quantitatively detected
626 metabolites (175 named, 19 isobars and 432 metabolites without
identification), of which 82.3% (515 out of 626) were shared by the
three diagnostic classes (Fig. 1b). Notably, there were 60 metabolites
found in PCA and/or metastatic tumours but not in benign prostate.
These profiles were displayed as a heat map (Fig. 1c) and z-score plot
(Fig. 1d). In the latter, benign-based z-scores were plotted for each of
the 626 metabolites. The plots revealed robust metabolic alterations in
metastatic tumours (z-score range: 213.6 to 81.9) compared to fewer
changes in clinically localized prostate cancer samples (z-score range:
27.7 to 45.8).
We identified the differential metabolites between the PCA and
benign samples using a two-sided Wilcoxon rank-sum test coupled
with a permutation test (n 5 1,000). A total of 87 out of 518 metabo-
lites were differential across these two classes (P , 0.05, corresponding
to a 23% FDR). For visualizing the relationship between the 87 altered
metabolites, hierarchical clustering was used to arrange the metabo-
lites on the basis of their relative levels across samples (Fig. 2a). Among
the perturbed metabolites, 50 were increased in PCA samples whereas
37 were downregulated. Figure 2b displays the relative levels of the 37
named metabolites that were differential between benign prostate and
PCA samples. Similarly, 124 out of 518 metabolites were found to be
increased in the metastatic samples compared to the localized
tumours and 102 compounds were downregulated (P , 0.05, corres-
ponding to a 4% FDR). Figure 2c displays the levels of the 91 named
metabolites altered in metastatic samples. A subset of six metabolites
including sarcosine, uracil, kynurenine, glycerol-3-phosphate, leucine
and proline were significantly increased on disease progression from
benign to PCA to metastatic prostate cancer. These metabolites could
potentially serve as biomarkers for progressive disease, one of the
factors that motivated us to examine sarcosine in greater detail.
NATURE | Vol 457 | 12 February 2009 LETTERS

a b c fits these criteria. Notably, metastatic samples showed markedly

n
Tissue PCA

ig

s
increased levels of sarcosine in 79% of the specimens analysed (chi-

A
n

et
(n = 42) (n = 12)

PC
Be

M
squared test, P 5 0.0538), whereas 42% of the PCA samples showed
225 3 an increase in the levels of this metabolite (Fig. 2a–c). Importantly,
none of the benign samples had detectable levels of sarcosine. Taken
105 120 32 18
176 515 together, this indicated the possible utility of sarcosine in monitoring
154 119 227 7 12 39 disease progression and aggressiveness.
To confirm this pattern of sarcosine increase in cancer progression,
Plasma Urine Benign Mets
we developed a highly sensitive and specific isotope dilution gas
(n = 110) (n = 110) (n = 16) (n = 14) chromatography–mass spectrometry (GC–MS) method for accurately
quantifying the metabolite from biospecimens (limit of detec-
tion 5 10 femtomoles, Supplementary Fig. 11). In an independent
d Benign PCA Mets set of 89 tissue samples (Supplementary Table 6), sarcosine levels were
significantly increased in PCA specimens (n 5 36) compared to benign
adjacent prostate samples (n 5 25, Wilcoxon P 5 4.34 3 10211,
600

600

600

Fig. 3a). Additionally, there was an even greater increase of sarcosine

in metastatic samples (n 5 28) compared to organ-confined disease
500

500

500

(Wilcoxon P 5 6.02 3 10211, Fig. 3a). In contrast, sarcosine was unde-

tectable in adjacent non-neoplastic tissues from patients with meta-
Metabolites

static disease (Supplementary Fig. 12a–c).

400

400

400
Metabolites

These findings led us to explore the potential of sarcosine as a

candidate for future development in biomarker panels for early disease
300

300

300

detection and aggressivity prediction. Towards this end, we monitored

its levels in urine specimens from biopsy-positive and -negative
200

200

200

individuals, most of whom have increased levels of prostate-specific

antigen (PSA) (.4.0 ng ml21) and in which prostate needle biopsy was
used for diagnosis. This is a particularly challenging cohort as, in
100

100

100

addition to these men being at high risk for prostate cancer, even a
negative needle biopsy does not rule out the presence of cancer due to
sampling issues. Sarcosine was found to be significantly higher in urine
0

–10 0 10 20 –10 0 10 20 –10 0 10 20
Deviation from benign
(standard deviations from average)
–2< –1 0 1 >2
Figure 1 | Metabolomic profiling of prostate cancer. a, Venn diagram of the
total metabolites detected across 42 prostate-related tissues and 110
matched plasma and urine samples. b, Venn diagram of 626 metabolites in
tissues measured across 16 benign adjacent prostate tissues, 12 clinically
localized prostate cancers (PCA) and 14 metastatic prostate cancers (Mets).
c, Heat map representation of unsupervised hierarchical clustering of the
data in b (rows) grouped by sample type (columns). Shades of yellow and
blue represent an increase and decrease of a metabolite, respectively, relative
to the median metabolite levels (see colour scale). d, z-score plots for the data
in b normalized to the mean of the benign prostate samples (truncated at 25
s.d. for clarity, see Supplementary Methods for procedural details).
Mapping the differential metabolomic profiles to their respective
biochemical pathways as outlined in the Kyoto Encyclopedia of Genes
and Genomes (KEGG, release 41.1, http://www.genome.jp/kegg,
Supplementary Fig. 8) revealed an increase in amino acid metabolism
and nitrogen breakdown pathways during cancer progression to meta-
static disease. A similar enrichment network of amino acid metabo-
lism was also identified by the bioinformatics tool Oncomine Concept
Map4,5(OCM, http://www.oncomine.org, P 5 6 3 10213, Supple-
mentary Figs 9 and 10a), supporting our earlier gene-expression-
based prediction of androgen-induced protein synthesis as an early
event during prostate cancer development5. Additionally, OCM found
strong enrichment for increased ‘methyltransferase activity’
(Supplementary Fig. 10b, P 5 7.7 3 1028) among metabolites upre-
gulated in metastatic samples. This corroborates previous studies from
our group and others showing an increase of the histone methyltrans-
ferase EZH2 in metastatic tumours6–11.
Because amino acid metabolism and methylation were enriched
during prostate cancer progression, we focused on differential meta-
bolites that characterize these processes and additionally show a pro-
gressive increase from benign to PCA to metastatic disease. The
amino acid metabolite sarcosine, an N-methyl derivative of glycine,
©2009 Macmillan Publishers Limited. All rights reserved
sediments (n 5 49, Wilcoxon P 5 0.0004, Fig. 3b) and supernatants
(n 5 59, Wilcoxon P 5 0.0025, Supplementary Fig. 14a) derived from
biopsy-positive prostate cancer patients compared to biopsy-negative
controls (n 5 44 and n 5 51, respectively; Supplementary Tables 7
and 8). The overall receiver operating characteristic curves for sarco-
sine indicate that its predictive value is modest, with an area under the
curve (AUC) of 0.71 for urine sediments and 0.67 for supernatants
(Supplementary Fig. 14b, c). Notably, an AUC of 1.0 indicates perfect
prediction and an AUC of 0.5 indicates prediction equivalent to
random selection. When restricted to samples having PSA in the
clinical grey zone of 2–10 ng ml21 (n 5 53), sarcosine performed
better than PSA in delineating the two diagnostic classes with an
AUC of 0.69 (95% CI: 0.55, 0.84) compared to an AUC of 0.53
(95% CI: 0.37, 0.69) for PSA (Supplementary Fig. 15). Thus, sarcosine
may have the potential to identify patients with modestly increased
PSA that are likely to have a positive prostate biopsy.
To determine whether the sarcosine increase in prostate cancer has
biological relevance, we measured its levels in prostate cancer cell lines
VCaP, DU145, 22RV1 and LNCaP (n 5 3 each) as well as in their
benign epithelial counterparts, primary benign prostate epithelial cells
(PrEC, n 5 2) and immortalized benign RWPE prostate cells (n 5 3).
Significantly increased levels of the metabolite were found in prostate
cancer cells compared to the benign cells (analysis of variance,
ANOVA, P 5 0.0218, Fig. 3c). Additionally, sarcosine levels correlated
well with cell invasiveness (Spearman’s correlation coefficient: 0.943,
P 5 0.0048). On the basis of our earlier findings that EZH2 overex-
pression in benign cells could mediate cell invasion and neoplastic
progression6,10,11, sarcosine levels were assessed on modulation of
EZH2 expression. Interestingly, overexpression of EZH2 in benign
prostate epithelial cells increased sarcosine levels (4.5-fold, Supple-
mentary Fig. 16a), whereas its knockdown in DU145 prostate
cancer cells diminished the levels of the metabolite (Supplementary
Fig. 16a–c). To determine whether sarcosine has a more direct role in
this process, we added the metabolite to non-invasive benign prostate
epithelial cells. Alanine, an isomer of sarcosine, was used as a control for
these experiments. Remarkably, the mere addition of exogenous sarco-
sine imparted an invasive phenotype to benign prostate epithelial cells
911
LETTERS NATURE | Vol 457 | 12 February 2009

a Benign PCA Mets c

ii ii iii iv v
Citrulline
p-Hydroxyphenyllactate
Cytidine β-Aminoisobutyrate
NAD 4-Acetamidobutanoate
Kynurenine
Homocysteine Thymine
Cysteine Myristate (14:0)
Ethylmalonate
Palmitoleate (16:1n7)
Uracil
Proline 2-Hydroxybutyrate (AHB)
Leucine Sarcosine
Sarcosine Palmitate (16:0)
Indoxylsulphate
Glycerol-3-phosphate
N-Acetylaspartate Laurate (12:0)
Sorbitol
Malate Erythritol
Glycine Glycerol
Threonine Linoleate (18:2n6)
Histidine
Asparagine Isobar 24 includes L-arabitol
Glutamate Uracil
2-Aminoadipate
Urea
Pipecolate
Kynurenine
Catechol
Sucrose/maltose
Kynurenine
N-Acetylglucosaminylamine Glycerol-3-phosphate
N-Acetylglucosamine Methylglutarate
N-Acetylgalactosamine 1-Methyladenosine
Uridine
Isobar includes glutamate Leucine
Bradykinin
Isobar includes 2-aminoisobutyrate
β-Alanine
Urate
Fructose Xanthine
Octadecanoic acid
Glucose Proline
2-Hydroxybutyrate (AHB)
Oleate (18:1n9)
Adenosine Glycocholate
Riboflavin (vitamin B2)
Glutathione reduced Glycerophosphorylcholine
Topiramate
Myo-inositol 5-Methylthioadenosine
Inosine Picolinate
Taurine S-Adenosylmethionine
Inositol-1-phosphate
Valine
Isobar includes mannose Tryptophan
Pantothenate
Arachidonate (20:4n6)
Heptadecanoate (margarate; 17:0)
Urate Fumarate (trans-butenedioate)
Creatinine Xanthosine
Homoserine lactone
Thyroxine
5-Hydroxyindoleacetate
–2< –1 0 1 >2 Isobar includes maltotetraose
Isobar includes gluconate
b Bradykinin
Caffeine
Cysteine DHEA-S
N-Acetylgalactosamine Fructose-6-phosphate
Sarcosine Argininosuccinate
Kynurenine Guanine
N-Acetylaspartate Mannose-6-phosphate
Uridine Adenosine
Glutamate Isobar includes arginine
Bradykinin Kynurenine
Uracil Cholesterol
Homocysteine N-Acetylglucosaminylamine
NAD Spermine
N-Acetylglucosamine g-Glutamylglutamine
Asparagine Spermidine
N-Acetylglucosaminylamine Serine
Glycine N-Acetylgalactosamine
2-Aminoadipate Isobar includes inositol 1-phosphate
Leucine N-Acetylglucosamine
Proline Aspartate
Glycerol-3-phosphate Hydroxyphenylpyruvate
Threonine Hexanoylcarnitine
Malate Taurine
Histidine Bradykinin hydroxyproline
Citrulline Uridine
Kynurenine Ribose
Cytidine Isobar includes D-saccharate
Isobar includes glutamate Citrate
Fructose 1,5-anhydroglucitol
Isobar includes mannose Phosphoserine
β-Alanine Nicotinamide
Glucose Ciliatine (2-aminoethylphosphonate)
Adenosine Myo-inositol
Inosine Putrescine
Urate N-6-trimethyllysine
Creatinine Inosine
Myo-inositol Guanosine
Taurine
Glutathione reduced (GSH)

–5 0 5 10 >15 –15 –10 –5 0 5 10 >15

Deviation from benign Deviation from PCA
(standard deviations from average) (standard deviations from average)

Figure 2 | Metabolomic alterations of prostate cancer progression. a, Heat metabolites from a. Each point represents one metabolite in one sample,
map showing 87 differential metabolites in PCA relative to benign samples coloured by tissue type (blue, benign; yellow, PCA). c, As in b except for the
(Wilcoxon P # 0.05). Localized PCA samples are grouped as (i) low grade comparison between Mets (red) and PCA (yellow), with data represented
(Gleason 313, 314) or (ii) high grade (Gleason 413, 414). Metastatic relative to the mean of the PCA samples. For clarity, the plots in b and c have
samples are grouped by the site of tissue procurement, namely (iii) soft been truncated at 15 standard deviations above the mean of the benign and
tissue, (iv) rib/diaphragm or (v) liver. b, Benign-based z-score plot of named PCA samples, respectively.
912
©2009 Macmillan Publishers Limited. All rights reserved
NATURE | Vol 457 | 12 February 2009 LETTERS

a b a b

Sarcosine concentration
Benign PCA Mets

(pmoles per 106 cells)

Concentration sarcosine
(pmoles per mg tissue) 500 2.0
60 2.02 e 1.8

(sarcosine/alanine)
2

Normalized log2
3.84 AR ERG GNMT Cancer 400 1.6

Invasion (A560)
50 HO HO CH3 Invasion 1.4
NH2 N
40 1 O O
H Migration 300 1.2
Aggressivity 1.0
Glycine Sarcosine
30 0 ERG SARDH 200 0.8
DMGDH 0.6
20 HO CH3 100 0.4
–1 N
10 0.2
O CH3 0 0.0
0 –2 Dimethylglycine

on trol

M NA t

A
G siR rge

RN
Biopsy- Biopsy-

on
-ta

si
negative positive

T
c

(relative to non-target siRNA)

d

(relative to non-target siRNA)

N
Fold change in invasion
Fold change in sarcosine
c d
Concentration sarcosine
(pmoles per 10 6 cells)

70 3.0 4 4
Alanine GNMT
Invasion (fold change
0.75 9 1.4
relative to control)
60

SARDH/GAPDH
Glycine SARDH

GNMT/GAPDH
3 3 8 1.2
Invasion (A560)

50 2.5 Sarcosine 7
6 0.8
40 0.50 2 2 5 0.6
2.0 4 1.0
30 1 1 3
0.25 2 0.4
20 1.5 0.2
0 0 1
10 0 0

A t

RN H
RN e
0 1.0 0 h 3 h 12 h 24 h 48 h

si RD
si arg

A
0 I Duration of androgen treatment
EC PE aP 145 RV aP

SA
-t
on
PR RW VC DU 22 LNC

N
10 25 50
Non- Invasive Amino acid concentration e g
(µM) Overexpression Knockdown
Chr6: 43,036| 43,040| (kb)
invasive

Sarcosine concentration
binding binding
RWPE VCaP

(pmoles per 106 cells)

1.00

ERG
40
Figure 3 | Sarcosine levels in prostate cancer and its association with cell 35
invasion. a, Sarcosine levels in prostate-cancer-related tissue specimens

Invasion (A560)
30 0.75

AR
(n 5 89). b, Sarcosine levels in post-digital-rectal-exam urine sediments 25
from men with biopsy-proven prostate cancer (n 5 49) and prostate-biopsy- GNMT 20 0.50
f Chr9: 135,595| 135,599| (kb) 15
negative controls (n 5 44). Asterisks indicate truncated measures. binding
10 0.25
ERG

c, Increased levels of sarcosine (black bars) were found in invasive prostate

5
cancer cells compared to non-invasive benign prostate epithelial cell lines. SARDH 0 0
Mean and s.e.m. of sarcosine levels (n 5 3, except for PrEC cells where

l
G
n- V1

NA
ro

ER iRN et
G A
ER
nt

s rg
ET
n 5 2). Cell invasion (grey bars) was also measured (mean and s.e.m.).

siR
Co

ta
d, Assessment of cell invasiveness of prostate epithelial cells upon exogenous

No
administration of alanine, glycine or sarcosine (mean and s.e.m., n 5 3).

(Fig. 3d and Supplementary Fig. 17). Furthermore, the number of Figure 4 | A role for sarcosine in androgen signalling and prostate cancer
motile prostate epithelial cells was significantly higher on sarcosine cell invasion. a, Schematic of the sarcosine pathway and its potential link to
prostate cancer. b, Assessment of sarcosine levels and cell invasiveness after
treatment (t-test P 5 6.997 3 1026, n 5 10) compared to alanine-
knockdown of GNMT in DU145 cells by RNA interference. c, As in b except
treated controls (Supplementary Fig. 18). Sarcosine treatment, however, knockdown of SARDH in RWPE cells (n 5 6). d, qRT–PCR analysis of
did not affect the ability of these cells to progress through the different GNMT and SARDH mRNA expression in androgen-stimulated VCaP cells.
stages of cell cycle (Supplementary Fig. 19a–d) or impair cell prolifera- e, Androgen receptor (AR) and ERG binding sites on the promoter of
tion (Supplementary Fig. 19e). Notably, glycine, a precursor of sarco- GNMT, as determined by ChIP-seq. The y axes display the number of reads
sine, induced invasion in these cells, although to a lesser degree than in a 25 bp sliding window. f, As in e, except ERG binding sites in the
sarcosine (Fig. 3d). This invasion could result from the conversion of promoter of SARDH. g, Left, overexpression of ERG or ETV1 in RWPE cells
glycine to sarcosine by the enzyme glycine-N-methyltransferase and measurement of sarcosine levels and cell invasiveness. Right, as in left,
(GNMT; Fig. 4a). except knockdown of TMPRSS2–ERG in VCaP cells by RNA interference. All
error bars represent mean and s.e.m., n 5 3 unless indicated otherwise.
In addition to GNMT, sarcosine levels are regulated by sarcosine
dehydrogenase (SARDH), the enzyme that converts sarcosine back to progression12, we investigated their role in regulating GNMT and
glycine, and dimethylglycine dehydrogenase (DMGDH), which SARDH. Treatment with androgen for 48 h in VCaP (ERG-positive)
generates sarcosine from dimethylglycine (Fig. 4a). By virtue of their and LNCaP (ETV1-positive) prostate cancer cells resulted in a step-
ability to control sarcosine levels in cells, these enzymes may assume a wise increase in GNMT expression and a concomitant decrease in
critical role in modulating prostate cancer invasion. To test this hypo- SARDH levels, as assessed by digital gene expression and quantitative
thesis, a series of RNA-interference-mediated knockdown experiments polymerase chain reaction (qPCR) (Fig. 4d and Supplementary
were carried out. Attenuation of GNMT (Fig. 4b, Supplementary Fig. 25e). This finding was supported by chromatin immunopreci-
Fig. 20) in DU145 prostate cancer cells resulted in a significant reduc- pitation sequencing (ChiP-Seq), which revealed direct binding of
tion in cell invasion (t-test P 5 0.0073, n 5 3), with a concomitant androgen receptor and ERG to the promoter of GNMT in VCaP cells
threefold decrease in the intracellular sarcosine levels compared to (Fig. 4e), whereas only ERG binding was seen on the SARDH
control non-target short interfering RNA (siRNA)-transfected cells. promoter (Fig. 4f). In ETV1-positive LNCaP cells, androgen recep-
Similar knockdown experiments performed in benign RWPE cells tor, but not ERG as expected, was bound to both GNMT and SARDH
significantly hampered the ability of exogenous glycine (t-test promoters (Supplementary Figs 25a, b). The binding data were
P 5 0.0082, n 5 3), but not sarcosine, to induce invasion (Supple- validated by ChIP–PCR (Supplementary Figs 24a, b, d and 25c, d),
mentary Fig. 21a, b). Comparable loss of cell invasion and reduction which additionally revealed weak binding of androgen receptor to the
in sarcosine levels were also apparent in DU145 cancer cells on knock- SARDH promoter in VCaP cells (Supplementary Fig. 24c). These
down of DMGDH (Supplementary Fig. 22a, b). In contrast, knock- findings together directly link activation of the sarcosine pathway
down of SARDH in benign prostate epithelial cells resulted in an to androgen receptor and ETS gene fusion regulation—two key med-
,3-fold increase in endogenous sarcosine levels with a concomitant iators of prostate cancer progression. Remarkably, both ERG- and
.3.5-fold increase in invasion (Fig. 4c and Supplementary Fig. 23). ETV1-induced invasion were associated with a threefold sarcosine
With the understanding that androgen signalling and ETS family increase in benign RWPE cells (Fig. 4g). Similarly, knockdown of the
of genes (ERG, ETV1) fusions are key factors for prostate cancer TMPRSS2–ERG gene fusion in VCaP cells (Supplementary Fig. 26)
913
©2009 Macmillan Publishers Limited. All rights reserved
LETTERS
resulted in a more than threefold decrease in sarcosine with a similar
decrease in the invasive phenotype (Fig. 4g).
Taken together, we explored the metabolome of prostate cancer
progression. This led to the characterization of metabolomic signa-
tures, which in the context of other molecular alterations may lead to
a more complete understanding of disease progression. Specifically,
we identified sarcosine as a key metabolite increased most robustly in
metastatic prostate cancer and detectable in the urine of men with
organ-confined disease. Interestingly, sarcosine and its proximal
regulatory enzymes seem to have an intermediary role in neoplastic
progression modulating cell invasion and migration. The master
transcriptional regulators of prostate cancer progression, androgen
receptor and the ETS gene fusions, seem to regulate directly sarcosine
levels by means of transcriptional control of its regulatory enzymes.
Thus, components of the sarcosine pathway may have potential as
biomarkers of prostate cancer progression and serve as new avenues
for therapeutic intervention.
METHODS SUMMARY
Biospecimens and associated clinical data related to the study were collected with
written consent from the University of Michigan and approved for use by the
Internal Review Board. Unbiased metabolomic profiling using liquid/gas chro-
matography coupled to mass spectrometry (LC/GC–MS) was performed as
described3 using a Thermofisher linear ion-trap mass spectrometer with Fourier
transform and Mat-95 XP mass spectrometers, respectively (Supplementary Fig. 1).
Target metabolites were assessed in tissue and urine samples using isotope dilution
GC–MS. Metabolomic data analysis is detailed in Supplementary Fig. 4. All
Wilcoxon rank-sum tests and t-tests are two-sided using a threshold of P , 0.05
for significance. Repeated measures ANOVA is used for the cell line data with
P values from the model F-test. Class-specific metabolomic patterns were visualized
using z-score plots and heat maps. Unsupervised clustering of samples using meta-
bolomic signatures was performed using Cluster13 and TreeView14, and visualized
using heat maps. Analysis of network relationships among various molecular con-
cepts and metabolomic data was performed using OCM4,5 (http://www.oncomine.
org), as outlined in Supplementary Fig. 9. Invasion was measured using a modified
Boyden chamber assay as described10. The cell motility assay was performed as
reported previously using blue fluorescent microsphere beads15. Targeted knock-
down of candidate genes16 using gene-specific siRNA sequences are listed in
Supplementary Table 9. qPCR for enzymes regulating sarcosine levels, EZH2 and
ETS, was performed as described12 using the indicated oligonucleotide primers
(Supplementary Table 10). Chromatin immunoprecipitation to interrogate the
regulatory role of androgen and ETS was performed using published protocols17.
ChIP-Seq and digital gene expression were measured using the Genomic DNA
Sample Prep Kit and the NIaIII kit on a Genome Analyser (Illumina) according
to the manufacturer’s instructions.
Full Methods and any associated references are available in the online version of
the paper at www.nature.com/nature.
Received 9 October 2008; accepted 6 January 2009.
1. Abate-Shen, C. & Shen, M. M. Molecular genetics of prostate cancer. Genes Dev.
14, 2410–2434 (2000).
2. Ruijter, E. et al. Molecular genetics and epidemiology of prostate carcinoma.
Endocr. Rev. 20, 22–45 (1999).
3. Lawton, K. A. et al. Analysis of the adult human plasma metabolome.
Pharmacogenomics 9, 383–397 (2008).
914
©2009 Macmillan Publishers Limited. All rights reserved
NATURE | Vol 457 | 12 February 2009
4. Rhodes, D. R. et al. Molecular concepts analysis links tumors, pathways,
mechanisms, and drugs. Neoplasia 9, 443–454 (2007).
5. Tomlins, S. A. et al. Integrative molecular concept modeling of prostate cancer
progression. Nature Genet. 39, 41–51 (2007).
6. Varambally, S. et al. The polycomb group protein EZH2 is involved in progression
of prostate cancer. Nature 419, 624–629 (2002).
7. van der Vlag, J. & Otte, A. P. Transcriptional repression mediated by the human
polycomb-group protein EED involves histone deacetylation. Nature Genet. 23,
474–478 (1999).
8. Laible, G. et al. Mammalian homologues of the Polycomb-group gene Enhancer of
zeste mediate gene silencing in Drosophila heterochromatin and at S. cerevisiae
telomeres. EMBO J. 16, 3219–3232 (1997).
9. Cao, R. et al. Role of histone H3 lysine 27 methylation in Polycomb-group
silencing. Science 298, 1039–1043 (2002).
10. Kleer, C. G. et al. EZH2 is a marker of aggressive breast cancer and promotes
neoplastic transformation of breast epithelial cells. Proc. Natl Acad. Sci. USA 100,
11606–11611 (2003).
11. Varambally, S. et al. Genomic loss of microRNA-101 leads to overexpression of
histone methyltransferase EZH2 in cancer. Science 322, 1695–1699 (2008).
12. Tomlins, S. A. et al. Recurrent fusion of TMPRSS2 and ETS transcription factor
genes in prostate cancer. Science 310, 644–648 (2005).
13. Eisen, M. B. & Brown, P. O. DNA arrays for analysis of gene expression. Methods
Enzymol. 303, 179–205 (1999).
14. Eisen, M. B., Spellman, P. T., Brown, P. O. & Botstein, D. Cluster analysis and
display of genome-wide expression patterns. Proc. Natl Acad. Sci. USA 95,
14863–14868 (1998).
15. Klemke, R. L. et al. Regulation of cell motility by mitogen-activated protein kinase.
J. Cell Biol. 137, 481–492 (1997).
16. Varambally, S. et al. The polycomb group protein EZH2 is involved in progression
of prostate cancer. Nature 419, 624–629 (2002).
17. Yu, J. et al. A polycomb repression signature in metastatic prostate cancer
predicts cancer outcome. Cancer Res. 67, 10657–10663 (2007).
18. Storey, J. D. A direct approach to false discovery rates. J. R. Stat. Soc. [Ser B] 64,
479–498 (2002).
19. Yu, J. et al. Integrative genomics analysis reveals silencing of beta-adrenergic
signaling by polycomb in prostate cancer. Cancer Cell 12, 419–431 (2007).
Supplementary Information is linked to the online version of the paper at
www.nature.com/nature.
Acknowledgements We thank J. Granger for help in manuscript preparation,
J. Siddiqui and R. Varambally for help with the clinical database, and A. Vellaichamy
and S. Pullela for technical assistance. We thank K. Pienta for access to metastatic
prostate cancer samples from the University of Michigan Prostate SPORE rapid
autopsy programme. This work is supported in part by the Early Detection
Research Network (A.M.C., J.T.W.), National Institutes of Health (A.S., S.P., J.B.,
T.M.R., D.G., G.S.O. and A.M.C.) and an MTTC grant (G.S.O. and A.S.). A.M.C. is
supported by a Clinical Translational Science Award from the Burroughs Welcome
Foundation. A.S. is supported by a grant from the Fund for Discovery of the
University of Michigan Comprehensive Cancer Center. L.M.P. is supported by the
University of Michigan Cancer Biostatistics Training Grant. A.M.C and S.P. are
supported by the Doris Duke Charitable Foundation.
Author Contributions A.S., L.M.P. and A.M.C. wrote the manuscript. A.S. and
A.M.C. conceptualized, designed and interpreted the data. L.M.P., R.J.L., S.K.-S.,
D.G. and D.C.A. performed data analysis. T.M.R., G.S.O., J.B. S.P., J.R.S., A.B. and
C.B. carried out the mass spectrometry studies. A.P.K., J.Y., Q.C., B.L., Y.L., M.K.N.,
A.A., X.C. and S.V. performed biochemical experiments. R.M., B.H., A.M.C. and
J.T.W. coordinated the clinical and pathology components of the study.
Author Information The authors declare competing financial interests: details
accompany the full-text HTML version of the paper at www.nature.com/nature.
Reprints and permissions information is available at www.nature.com/reprints.
Correspondence and requests for materials should be addressed to A.M.C.
(arul@umich.edu).
doi:10.1038/nature07762
METHODS
Biospecimens and cell lines. Prostate tissues, urine and plasma were obtained
from the University of Michigan SPORE and EDRN Tissue Core. All samples
were collected with informed consent as per the approval of the Institutional
Review Board.
RWPE, DU145, LnCAP and PC3 cells were obtained from ATCC, PrEC cells
from Cambrex BioScience, 22-RV1 was provided by J. Macoska and VCaP by K.
Pienta. VCaP and LnCAP were grown in charcoal-stripped-serum-containing
media for 24 h, before treatment for a further 24 h with vehicle or 1 nM methyl-
trienolone (R1881, NEN) dissolved in ethanol.
Metabolomic profiling. Metabolomic profiling was performed using the plat-
form described previously3 and outlined in Supplementary Fig. 1. The LC–MS
portion of the platform is based on a Surveyor HPLC and a Thermo-Finnigan
LTQ-FT mass spectrometer (Thermo Fisher Corporation) with the instrument set
for continuous monitoring of both positive and negative ions. Samples that were
analysed by GC–MS were derivatized under dried nitrogen using bistrimethyl-
silyl-triflouroacetamide (BSTFA) and analysed on a Thermo-Finnigan Mat-95 XP
using electron impact ionization and high resolution. For both LC–MS and GC–
MS, spectral files were searched using metabolomic libraries created by Metabolon
that contain about 800 commercially available compounds.
Quantification of target metabolites was performed by isotope dilution GC–
MS using selected ion monitoring (SIM). The samples were modified to their
t-butyl dimethylsilyl derivatives and analysed with an Agilent 5975 MSD mass
detector using electron impact ionization. For SIM analysis, the m/z for native
and labelled molecular peaks for various target metabolites quantified were: 158
and 161 (sarcosine), 406 and 407 (cysteine), 432 and 437 (glutamic acid), 297
and 301 (thymine) and 218 and 219 (glycine), respectively. Assessment of citric
acid was performed on the GC–MS in the full scan mode.
Statistical analysis. Missing metabolite measurements were replaced (imputed)
with zero for metabolites where the mean relative standardized intensity measure
was over 100,000 across the samples, that is, we assume missingness was probably
due to absence of the metabolite in the sample. Otherwise one half of the sample
minimum was used to replace the missing measurement, that is, we assume miss-
ingness was probably due to detection limits. Imputed data were median-centred
and inter-quartile-range-scaled per sample. Plotted z-scores were calculated based
on the mean and standard deviation of a reference set (benign samples, unless
otherwise stated). Hierarchical clustering14 based on Pearson’s correlation was
performed on the log-transformed normalized data after median centring per
metabolite. A small value (unity) was added to each normalized value to allow
log transformation. Per-metabolite chi-squared tests were used to assess class-
specific metabolite patterns of present and absent (undetected) measurements.
Per-metabolite two-tailed Wilcoxon rank sum tests were used for two-sample tests
of association between classes. Kruskal–Wallis tests were used for three-way com-
parisons between all diagnosis groups. Non-parametric tests were chosen to reduce
the influence of the imputed values. Tests were run on those metabolites with
detectable expression in at least 20% of the samples. Significance was determined
©2009 Macmillan Publishers Limited. All rights reserved
using 1,000 sample permutations. FDRs were calculated using the q-value conversion
algorithm of ref. 18. Pairwise differences in expression in the cell line data and small-
scale tissue data were tested using two-tailed t-tests with Satterthwaite variance
estimation. Comparisons involving multiple cell lines used repeated measures
ANOVA to adjust for the multiple measures per cell line. Fold change was estimated
using ANOVA on a log scale, following the model log(Y) 5 A 1 B*treatment 1 E. In
this way exp(B) is an estimate of (Y j treatment 5 1)/(Y j treatment 5 0), and the
standard error of exp(B) can be estimated from SE(B) using the delta method. The
threshold for significance was P , 0.05 for all tests.
ChIP-PCR. ChIP was carried out as described previously19 using antibodies against
androgen receptor (Millipore), ERG (Santa Cruz) and rabbit IgG (Santa Cruz).
Androgen receptor ChIP was performed in paired ethanol-treated and R1881-
treated samples. ChIP-enriched chromatin as well as the whole-cell extract was
amplified by ligation-mediated PCR. When examining androgen receptor binding
on target genomic regions, equal amounts of ethanol-treated and R1881-treated
ChIP amplicons were subjected to qPCR, and the fold enrichment (R1881/ethanol)
was determined based on the cycle differences after normalization to input DNA.
For ERG ChIP assays, VCaP cells grown in regular medium were used for ChIP
using antibodies against ERG and rabbit IgG control. ChIP products were directly
analysed by the qPCR assay, and ERG binding was evaluated based on the cycle
difference between ChIP-enriched chromatin by ERG and corresponding IgG. The
primers used are listed in Supplementary Table 9.
ChIP-Seq. ChIP samples were prepared for sequencing using the Genomic DNA
Sample Prep kit following the manufacturer’s protocols. ChIP-sequencing was
performed using Illumina Genome Analyser according to standard manufac-
turer’s procedures. The raw sequencing image data were analysed by our analysis
pipeline and aligned to the unmasked human reference genome (NCBI v36,
hg18) using ELAND software to generate sequence reads of 25–32 base pairs.
Digital gene expression analysis. Trizol-extracted RNA from samples with 0 h
and 48 h androgen treatment was prepared for sequencing using the Digital Gene
Expression-Tag Profiling with NIaIII kit (Illumina) and sequenced by the
Genome Analyser. Sequencing reads were mapped back to the human reference
genome using the ELAND software. The number of sequencing reads for genes of
interest was counted. The expression level of each gene was measured as the
number of transcripts per million of total sequencing reads.
Quantitative RT–PCR. qPCR was performed using SYBR Green Mastermix on
an Applied Biosystems 7300 PCR machine as described previously19. All primers
were designed using Primer 3 and synthesized by Integrated DNA Technologies,
and are listed in Supplementary Table 9.
RNA interference. DU145 or RWPE cells were treated with non-targeting siRNA
(Dharmacon) and gene-specific siRNA sequences as listed in Supplementary
Table 10.
Cell invasion assay. Cell invasion assays were carried out using a modified
basement membrane chamber assay as described previously19.
Cell motility assay. For cell motility assays, we used the Cellomics cell motility
kit as per manufacturer’s instructions.